• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.6
• /
• pp.75-83
• /
• 2007

A model based approach of tracking the limbs of a walking human subject is proposed in this paper. The tracking process begins by building a data base composed of conditional probabilities of motions between the limbs of a walking subject. With a suitable amount of video footage from various human subjects included in the database, a probabilistic model characterizing the relationships between motions of limbs is developed. The motion tracking of a test subject begins with identifying and tracking limbs from the surveillance video image using the edge and silhouette detection methods. When occlusion occurs in any of the limbs being tracked, the approach uses the probabilistic motion model in conjunction with the minimum cost based edge and silhouette tracking model to determine the motion of the limb occluded in the image. The method has shown promising results of tracking occluded limbs in the validation tests.

• Journal of Korea Society of Industrial Information Systems
• /
• v.15 no.2
• /
• pp.91-98
• /
• 2010

This study proposes a monitoring system that effectively watches surroundings by cooperating the various sensor information including image information on a sensor network system. The monitoring system proposed in this paper is developed to watch certain intruders to the internal spaces through the interested region for exceptional time by installing cameras, PIR(Pyroelectric Infrared Ray) sensor and body detectors in such interested regions. Moreover the monitering system is implemented based on the SenWeaver plateform which is a integrated development tools for building wireless sensor network system. In the results of the test that was applied to a practically experimental environment by implementing some interfaces for the proposed system, it was considered that it is possible to watch surroundings effectively using the image information obtained from cameras and multiple sensor information acquisited from sensor nodes.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.4
• /
• pp.47-57
• /
• 2009

Event detection and recognition is an active and challenging topic recent in Computer Vision. This paper describes a new method for recognizing events caused by human motion from video sequences in an office environment. The proposed approach analyzes human motions using Motion History Image (MHI) sequences, and is invariant to body shapes. types or colors of clothes and positions of target objects. The proposed method has two advantages; one is thant the proposed method is less sensitive to illumination changes comparing with the method using color information of objects of interest, and the other is scale invariance comparing with the method using a prior knowledge like appearances or shapes of objects of interest. Combined with edge detection, geometrical characteristics of the human shape in the MHI sequences are considered as the features. An advantage of the proposed method is that the event detection framework is easy to extend by inserting the descriptions of events. In addition, the proposed method is the core technology for event detection systems based on context-aware computing as well as surveillance systems based on computer vision techniques.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.48 no.9
• /
• pp.22-30
• /
• 2011

Recently, the rising demand for intelligent video surveillance system leads to high-performance face recognition systems. The solution for low-resolution images acquired by a long-distance camera is required to overcome the distance limits of the existing face recognition systems. For that reason, this paper proposes a hardware design of an image resolution enhancement algorithm for real-time intelligent video surveillance systems. The algorithm is synthesizing a high-resolution face image from an input low-resolution image, with the help of a large collection of other high-resolution face images, called training set. When we checked the performance of the algorithm at 32bit RISC micro-processor, the entire operation took about 25 sec, which is inappropriate for real-time target applications. Based on the result, we implemented the hardware module and verified it using Xilinx Virtex-4 and ARM9-based embedded processor(S3C2440A). The designed hardware can complete the whole operation within 33 msec, so it can deal with 30 frames per second. We expect that the proposed hardware could be one of the solutions not only for real-time processing at the embedded environment, but also for an easy integration with existing face recognition system.

• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.2
• /
• pp.29-38
• /
• 2021

The Internet including mobile networks is developing to overcoming the limitation of physical distance and providing or acquiring information from remote locations. However, the systems that use video as primary information require higher bandwidth for recognizing the situation in remote places more accurately through high-quality video as well as lower latency for faster interaction between devices and users. The emergence of the 5th generation mobile network provides features such as high bandwidth and precise location recognition that were not experienced in previous-generation technologies. In addition, the Mobile Edge Computing that minimizes network latency in the mobile network requires a change in the traditional system architecture that was composed of the existing smart device and high availability server system. However, even with 5G and MEC, since there is a limit to overcome the mobile network state fluctuations only by enhancing the network infrastructure, this study proposes a high-definition video streaming system in ultra-low latency based on the SRT protocol that provides Forward Error Correction and Fast Retransmission. The proposed system shows how to deploy software components that are developed in consideration of the nature of 5G and MEC to achieve sub-1 second latency for 4K real-time video streaming. In the last of this paper, we analyze the most significant factor in the entire video transmission process to achieve the lowest possible latency.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.18 no.5
• /
• pp.171-182
• /
• 2019

Bicycles are a pollution-free means of transportation. In addition to leisure, the use of bicycles is increasing as alternative eco-friendly transportation. Accordingly, bicycle accidents are also increasing. The purpose of this study is to implement bicycle black box technology to identify situation when a bicycle accident occurs. Currently, bicycle black box products are mainly based on video cameras, and are commercially available by adding various functions mainly on high resolution cameras and are sold at high prices. If a bicycle accident occurs, quantitative data on the accident location at the time of the accident and the state of the bicycle at the time of the accident is required. In this study, IMU sensor used to obtain acceleration and slope, and time and coordinates are obtained. In addition, real-time acceleration and tilt data while is stored in memory card and by using Bluetooth transmit to the smart phone owned by the in real time to prevent accidents and to monitor status.

• Korean Chemical Engineering Research
• /
• v.57 no.2
• /
• pp.274-282
• /
• 2019

In recent years, big data analysis has been expanded to include automatic control through reinforcement learning as well as prediction through modeling. Research on the utilization of image data is actively carried out in various industrial fields such as chemical, manufacturing, agriculture, and bio-industry. In this paper, we applied NASNet, which is an AutoML reinforced learning algorithm, to DeepU-Net neural network that modified U-Net to improve image semantic segmentation performance. We used BRATS2015 MRI data for performance verification. Simulation results show that DeepU-Net has more performance than the U-Net neural network. In order to improve the image segmentation performance, remove dropouts that are typically applied to neural networks, when the number of kernels and filters obtained through reinforcement learning in DeepU-Net was selected as a hyperparameter of neural network. The results show that the training accuracy is 0.5% and the verification accuracy is 0.3% better than DeepU-Net. The results of this study can be applied to various fields such as MRI brain imaging diagnosis, thermal imaging camera abnormality diagnosis, Nondestructive inspection diagnosis, chemical leakage monitoring, and monitoring forest fire through CCTV.

• Journal of Institute of Control, Robotics and Systems
• /
• v.19 no.3
• /
• pp.202-208
• /
• 2013

In the image processing for VS (Video Surveillance), the detection of moving entities in a monitored scene is an important step. A background subtraction technique has been widely employed to find the moving entities. However, the extracted foreground regions often include not only real entities but also their cast shadows, and this can cause errors in following image processing steps, such as tracking, recognition, and analysis. In this paper, a novel technique is proposed to determine the shadow pixels of moving objects in the foreground image of a VS camera. Compared to existing techniques where the same decision criteria are applied to all moving pixels, the proposed technique determines shadow pixels using local features based on two facts: First, the amount of pixel intensity drop due to a shadow depends on the intensity level of background. Second, the distribution pattern of pixel intensities remains even if a shadow is cast. The proposed method has been tested at various situations with different backgrounds and moving humans in different colors.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.17 no.10
• /
• pp.2419-2424
• /
• 2013

In a conventional MOG algorithm, a small threshold for background decision causes the background recognition delay in a periodic background and a large threshold makes it recognize passing objects as background in a stationary background. This paper proposes the improved MOG algorithm using adaptive threshold. The proposed algorithm estimates changes of weight in the dominant model of the MOG algorithm both in the short and long terms, classifies backgrounds into the stationary and periodic ones, and assigns proper thresholds to them. The simulation results show that the proposed algorithm decreases the maximum number of frame in background recognition delay from 137 to 4 in the periodic background keeping the equal performance with the conventional algorithm in the stationary background.

• The Journal of the Korea Contents Association
• /
• v.3 no.1
• /
• pp.71-79
• /
• 2003

Up to row, researchers have observed experiment environment to study about raising procedure and an ecology of the ocean fishes with naked eye. Then they can monitor status information in only laboratory such as temperatures, image of the fish farm, etc. Such the existing way as this can't observe an ecology of the ocean fishes with scientific and systematic method and moreover it needs many of researchers. So in this paper, we design and implement a fish farm monitoring system which receives status of remote experiment environment through the internet whenever you want it and wherever you want it, and monitors and analyzes its status using embedded system of which the operating system is linux.